1
How to Save Costs Beyond Targeting the
Most Costly Patients?
Session # 278, February 14, 2019
Nils Fischer, Senior Analyst, Partners Healthcare Pivot Labs
Mariana Simons, Senior Data Scientist, Philips Research
2
Nils Fischer, MPH
Has no real or apparent conflicts of interest to report.
Mariana Simons, PhD
Salary: Philips Research employee
Receipt of Intellectual Property Rights/Patent Holder: Yes
Conflict of Interest
3
Agenda
4
Learning Objectives
1. Describe how EHR and PERS* data integration can
provide insights into the patient's health status
2. Analyze the differences between the characteristics of
patients in the Bottom-, Middle- and Top-cost segments
3. Compare dynamics of annual longitudinal trends of inpatient
costs in the different cost segments
4. Apply predictive analytics & tailored interventions in clinical
practices
* PERS - personal emergency response service
5
Partners HealthCare
6
Poll relating to healthcare challenge
1. In the United States, the net Medicare spending
will change from 3.5% of the gross domestic
product (GDP) in 2014 to what percentage in
2039?
A. 15.7%
B. 8.1%
C. 5.7%
D. 1.5%
https://live.eventbase.com/polls?event=himss19&polls=5125
Congress of the United States Congressional Budget Office. 2016. [2018-04-02]. The 2014 Long-Term
Budget https://www.cbo.gov/sites/default/files/113th-congress-2013-2014/reports/45471-long-termbudgetoutlook7-29.pdf
7
Poll relating to healthcare challenge
2. In the United States, what % of Medicare
patients accounted for nearly 50% of healthcare
costs?
A. 50%
B. 30%
C. 10%
D. 5%
Long-term trends in the concentration of Medicare spending. Riley GF. Health Aff (Millwood). 2007 May-Jun; 26(3):808-16.
https://live.eventbase.com/polls?event=himss19&polls=5126
8
Poll relating to healthcare challenge
3. What proportion of all hospital admissions were
potentially avoidable according to Centers for
Medicare & Medicaid Services (CMS)?
A. 10%
B. 25%
C. 50%
D. 66%
Segal M, Rollins E, Hodges K, Roozeboom M. Medicare-Medicaid eligible beneficiaries and potentially avoidable hospitalizations. Medicare Medicaid
Res Rev. 2014;4(1) doi: 10.5600/mmrr.004.01.b01
https://live.eventbase.com/polls?event=himss19&polls=5127
9
Partners Healthcare Pivot Labs and Philips
A collaboration to tackle healthcare’s biggest challenges…
Challenge 1
Is there untapped potential to control
costs and improve care beyond the
focus on highest-cost patients?
Challenge 2
How well predictive analytics can
identify patients at-risk of ER visits
so that early intervention can lower
overall healthcare costs?
10
Alice’s Journey
79 years, female
Widow, living alone
CHF, 6 years
2 hospitalizations last year
Personal Emergency
Response Service (PERS),
1 year
11
Incidents Transport to ER ER visit In-/Out-patient stay
The power of integrated data
Alice’s adverse events last year
* - outside of the Partners HCO patient leakage
1 2 3 4
Learning objectives
Alice’s Journey
Learning objective: 1. Describe how EHR and PERS data integration can
provide insights into the patient's health status
PERS service
Healthcare organization (HCO)
% hospitalized
55 33 33
1 out-patient
1 in-patient
1 in-patient*
1 out-patient
1 in-patient
1 in-patient*
12
Patients like Alice
A retrospective look*
Objectives
Evaluate healthcare utilization of patients using PERS
Analyze their healthcare cost
Patient sample
2,624 patients identified from Partners Healthcare and Lifeline
Data sources
A combination of Partners’ EHR data and Lifeline PERS data in 2011-2015
Patient
PERS service
Clinicians
* Agboola S, Golas S, Fischer N, Simons M, op den Buijs J, Schertzer L, Kvedar J, Jethwani K. Healthcare Utilization in Older Patients Using
Personal Emergency Response Systems: An Analysis of Electronic Health Records and Medical Alert Data, BMC Health Services Research,
2017, 17:282, DOI : 10.1186/s12913-017-2196-1
13
Study Population Profile
≥ High school educated
Married/Partnered
White
Male
Aged >80
42%
26%
80%
24%
60%
%
PATIENTS
# MEDICAL
COND.
23% 1
25% 2
21% 3
31% ≥4
Top 3
Inpatient Principal Diagnosis Groups
Overall (2011-2015)
n = 5,258 (%)
Congestive
Heart Failure (CHF) 5.7%
Chronic
Obstructive Pulmonary Disease (COPD) 4.6%
Dysrhythmias
4.3%
Inpatient Admissions
related to chronic conditions 52%
Potentially avoidable admissions*
37%
2,624 patients
* - Segal M, Rollins E, Hodges K, Roozeboom M. Medicare-Medicaid Eligible Beneficiaries and Potentially Avoidable Hospitalizations.
Medicare Medicaid Res Rev. 2014;4(1):1-13. doi:10.5600/mmrr.004.01.b01.
14
Costs of the M-segment patients are highest, increase fastest*
Healthcare Cost Increase
1 2 3 4
Learning objectives
p=0.003
p=0.002
p=0.008
p=0.003
5%
6-50%
51-100%
T
B
M
Cost segmentation
*Agboola S, Simons M, Golas S, op den Buijs J, Felsted J, Fischer N, Schertzer L, Orenstein A, Jethwani K, Kvedar J, Health Care Cost Analyses for
Exploring Cost Savings Opportunities in Older Patients: Longitudinal Retrospective Study, JMIR Aging 2018;1(2):e10254) doi:10.2196/10254
Learning objective: 2. Analyze the differences between the
characteristics of patients in the B-, M- and T-cost segments
15
Costly CHF admissions increased significantly in the M-segment
CHF-related hospitalizations
1 2 3 4
Learning objectives
p=0.096 (>0.05) in Tseg is
not statistically significant
16
M-segment has 75% of potentially avoidable hospitalizations
1- Segal M, Rollins E, Hodges K, Roozeboom M. Medicare-Medicaid Eligible Beneficiaries and Potentially Avoidable Hospitalizations.
Medicare Medicaid Res Rev. 2014;4(1):1-13. doi:10.5600/mmrr.004.01.b01.
Avoidable Hospitalizations
1 2 3 4
Learning objectives
Learning objective: 3. Compare dynamics of annual longitudinal trends of
inpatient costs in the different cost segments
17
M-segment has 60% of potential cost savings
Avoidable Costs
1 2 3 4
Learning objectives
18
Takeaway
Patients in the M-segment
offer the greatest potential
opportunity to avoid
hospitalizations and
reduce costs.
AI is needed to identify M-
segment patients since
there are 9 times more
patients in the M-segment
compared to the T-
segment.
19
Patients like Alice
A prospective look*
Objective - Evaluate the impact of predictive
analytics and tailored interventions on
healthcare utilization of M-segment patients.
Study
Population
370
patients
Intervention
Group
Intervention
Group
Control
Group
Control
Group
Observation
Period
(3 months)
Control
(care as usual)
Control
(care as usual)
Intervention
(care as usual)
Intervention
(care as usual)
Intervention
(tailored
interventions)
Intervention
(tailored
interventions)
Intervention
Period
(6 months)
Control
(care as usual)
Control
(care as usual)
* - RCT on Evaluating the Impact of an Integrated Risk Assessment System (Lifeline Personal Emergency Response Service) on
Healthcare Utilization, ClinicalTrials.gov, https://www.clinicaltrials.gov/ct2/show/NCT03126565.
20
The power of predictive analytics* and
tailored interventions
1 2 3 4
Learning objectives
Avoidable
Hospitalizations &
Costs
1. Predict
2. Intervene
3. Reduce
* op den Buijs J, Simons M, Golas S, Fischer N, Felsted J, Schertzer L, Agboola S, Kvedar J, Jethwani K, Predictive Modeling of 30-Day Emergency
Hospital Transport of Patients Using a Personal Emergency Response System: Prognostic Retrospective Study, JMIR Med Inform 2018;6(4):e49,
doi:10.2196/medinform.9907
Learning objective: 4. Apply predictive analytics & tailored interventions in
clinical practices
21
Predictive
algorithm
calculates
Needs identified and if needed
clinicians input requested
Tailored intervention, e.g. educational
support over next 4 weeks
RISK SCORE
HIGH RISK
Nurse triage based on Needs
Assessment Questionnaire
LOW RISK
_
_
Reassess
185 patients
The Stepped-Care Approach*
Intervention Group only
* Palacholla R, Fischer N, Agboola S, Simons M, Odametey S, Golas S, op den Buijs J, Schertzer L, Kvedar J, Jethwani K, Evaluating the Impact of a
Web-Based Risk Assessment System (CareSage) and Tailored Interventions on Health Care Utilization: Protocol for a Randomized Controlled Trial,
JMIR Res Protoc 2018;7(5):e10045, doi:10.2196/10045
22
Primary - 90- and 180- day ER visits
Secondary
Total healthcare cost (billing and internal cost to the hospitals)
Hospital admissions (+ avoidable as defined by CMS)
30, 90, 180-day readmissions
180-day mortality rate
Study Outcomes
23
Alice’s Risk Assessment
and Intervention Plan
Intervention Plan
o PCP consultation about
medication reconciliation
o Telephone education session
Medication adherence
Symptoms recognition
Diet & Physical activity
Alice’s journey (hypothesis) : ER visits and hospitalizations will be
reduced and Alice’s move from M- to T-segment will be prevented.
3.5
2.4
1.9
Month 1 Month 2 Month 3
Risk Score (0
-6)
Risk of Emergency Transport
24
Takeaways
Cost acuity pyramid
Present Future
5%
6-50%
51-100%
T
B
M
M-seg had the fastest increase in healthcare
costs.
M-seg had the fastest increase in conditions
that lead to avoidable hospital admissions.
M-seg had 75% of all avoidable admissions
and 60% of associated cost savings.
To maximize cost savings, HCOs
and Payers would benefit
focusing on M-segment in
addition to T-segment patients.
Population Health Management based on predictive analytics
& tailored interventions will enable clinicians to proactively
manage patients in their home or community environments,
beyond institutional settings.
25
Nils Fischer, nfischer@partners.org
Mariana Simons, mariana.simons@philips.com
Please complete the online session evaluation
Questions